2026-04-16
Anthropic releases Claude Opus 4.7 with improved coding and 3.75 MP vision
Claude Opus 4.7 ships to Claude products, API, Bedrock, Vertex, and Microsoft Foundry. Better coding, ~3x higher vision resolution, same pricing.
Anthropic released Claude Opus 4.7 on April 16. Per the announcement, the model is available in Claude products, on the API as claude-opus-4-7, and through Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. Headline improvements: better performance on complex coding and long-running agentic tasks, vision now handles up to ~3.75 MP images (about 3x prior resolution), and pricing is held at $5/M input and $25/M output.
Anthropic also notes the updated tokenizer can increase token usage by 1.0–1.35x relative to prior Opus, and that “instructions are now followed more literally.”
Practitioner note
If you’ve tuned prompts on Opus 4.6, the “more literal instruction following” note is the one I’d re-validate first — soft instructions (“be concise”) that previously moderated tone may now be applied more strictly. Worth a small eval pass on your existing system prompts before flipping production traffic.
The vision resolution bump is more useful than it sounds: 3.75 MP unlocks readable text in screenshots and architectural diagrams without needing pre-resize.